arrow drop search cross

Séminaire de Damien Querlioz

Physics-based computing: the brain-inspired approach


15/03/2018 Nice Valrose
Publication : 15/03/2018
Partager cet article :

Damien Querlioz, C2N, Université Paris Sud

In recent years, Artificial Intelligence algorithms have made considerable progress, but their performance comes at a considerable cost: energy. Brains, by contrast perform intelligent tasks with extreme energy efficiency, by exploiting the biophysics of basic nanostructures such as synapses and ion channels. In order to approach brains efficiency, many electrical engineers and physicists are now designing systems that attempt to use electronic devices a lot more like brains are using their own basic devices. In the first part of my talk, I will show two examples of such projects:  a speech recognition system that uses magnetic nanooscillators as neurons, and a learning system that use nanodevice-based memories as synapses.  

We will see that these approaches of computing using device physics have some parallels with quantum computing, but also strong differences. In the final part of my talk, I will try to express questions raised by the comparison of quantum and brain-inspired computing.